A superconducting qubit is like a finicky musical instrument: until you tune it, you won't get a clean note. But even after tuning, the pitch drifts. The culprits are tiny defects in the material that physicists call two-level systems. They're like invisible cracks that appear and disappear, throwing off the processor's rhythm.
The researchers built a setup based on a field-programmable gate array (FPGA)—a flexible microchip that can be reprogrammed for a task. With it, they measured qubit relaxation with sub-second resolution, catching changes faster than a second. It turned out the picture is the same across different qubits, different labs, and different fabrication processes. Moreover, these fast jumps were directly linked to errors in quantum gates — the basic operations that make up a computation. The link was confirmed using randomized benchmarking: a method that estimates gate fidelity by randomly sampling sequences of gates.
Now it's clear: tuning a quantum computer is not a one-time affair but a continuous battle against fast noise. Returning to the musical analogy: the instrument goes out of tune not in a month but in a couple of seconds, and the tuner has to work nonstop.
Two-level systems are not an abstraction but real atoms or groups of atoms that can exist in two states, like tiny switches. Their switching saps energy from the qubit, causing it to "forget" its quantum state.
In Neal Stephenson's novel The Diamond Age, nanomachines battle material defects at the atomic level — much like engineers must fight two-level systems in quantum processors.